Proton sponge or membrane fusion? Endosomal escape of siRNA polyplexes illuminated by molecular dynamics simulations

This study utilizes All Atom and Coarse Grained Molecular Dynamics simulations to elucidate and differentiate the endosomal escape mechanisms of various siRNA polyplexes and lipid nanoparticles, revealing the critical roles of hydrophobic residues and anionic lipids while successfully correlating these interaction patterns with in vitro performance.

The Gist

The Big Picture: The "Locked Box" Problem

Imagine you want to send a secret message (a drug called siRNA) to a specific room inside a city (your cells) to fix a broken machine (a disease gene).

1. The Delivery: You put the message inside a protective delivery truck (a nanoparticle).
2. The Trap: The city guards (the cell) swallow the truck whole, but they don't let it into the main city hall. Instead, they lock it inside a small, temporary holding cell called an endosome.
3. The Goal: For the medicine to work, the truck must escape the holding cell before it gets crushed and recycled. This is called Endosomal Escape.

For years, scientists have been trying to build trucks that can break out of these holding cells, but they haven't fully understood how the best trucks do it. This paper uses powerful computer simulations to watch these trucks try to break out in real-time, revealing the secrets of success.

---

The Cast of Characters

The researchers tested five different types of delivery trucks to see which one was the best escape artist:

1. The Classic Sponge (bPEI): An old-school polymer that acts like a sponge. It soaks up acid, swells up, and hopes to burst the cell open.
2. The Hybrid Truck (PPP): A mix of the classic sponge and a fatty chain. It's a bit of a middle-ground.
3. The Wet vs. Dry PBAE: Two versions of a new polymer.
   • 30% OA (The Wet One): More water-loving, less greasy.
   • 70% OA (The Greasy One): Packed with oily, hydrophobic tails.
4. The Gold Standard (LNP): The "Onpattro" truck. This is the only one currently approved by the FDA for human use. It's the benchmark everyone tries to beat.

---

The Experiments: What Happened in the Lab?

First, the team tested these trucks in actual human cells (HeLa cells).

• The Winner: The Greasy PBAE (70% OA) and the Gold Standard LNP were the only ones that successfully delivered the message and turned off the disease gene.
• The Losers: The Classic Sponge and the Wet PBAE barely got the message out. They stayed trapped in the holding cell.
• The Side Effect: The Greasy PBAE was very effective, but it was also a bit violent. It tore holes in the holding cells so big that it sometimes hurt the cell itself (toxicity). The LNP was effective but gentler.

The Mystery: Why did the Greasy PBAE work so well? Was it just because it was "stronger"? Or was there a specific trick it used?

---

The Computer Simulation: The "Fly on the Wall"

Since we can't see inside a cell with a microscope well enough to see molecules moving, the researchers used Molecular Dynamics (MD) simulations. Think of this as a super-advanced video game where they built a digital model of the cell membrane and the trucks, then hit "play" to watch what happens over a few microseconds.

They used two levels of detail:

• Coarse Grained (CG): Like looking at the trucks from a drone. You see the shapes and how they bump into things, but not every single atom. This lets them watch for a long time.
• All Atom (AA): Like zooming in with a microscope to see every single atom. This shows the tiny details of how they stick together.

The Big Discovery: "The Grease Factor"

The simulations revealed two very different ways the trucks tried to escape:

1. The "Sponge" Strategy (Classic bPEI):
The Classic Sponge truck tried to use the "Proton Sponge" theory. It soaked up acid, swelled up, and pushed against the cell wall like a balloon trying to pop a balloon.

• Result: It barely made a dent. It stuck to the surface but couldn't get inside the wall. It was like trying to break a brick wall by blowing on it.

2. The "Grease" Strategy (Greasy PBAE & LNP):
The trucks with the oily (hydrophobic) tails did something completely different.

• The Analogy: Imagine the cell membrane is a layer of oil and water. The Greasy trucks didn't just push; they dove in. Their oily tails reached out and grabbed the oily tails of the cell membrane.
• The Fusion: Instead of just pushing, the truck and the cell wall started to mix. The truck's oil merged with the cell's oil. This created a hole or a "fusion" where the message could slip through.
• The Result: This was much more effective. The Greasy PBAE and the LNP both used this "merging" technique to break out.

The Role of the "Velcro":
The simulations also showed that the cell membrane has "Velcro" patches (negatively charged lipids). All the trucks were attracted to these patches. However, the Greasy trucks used the Velcro to stick close, and then used their oily tails to melt into the wall. The non-greasy trucks stuck to the Velcro but couldn't melt in, so they stayed trapped.

---

The Takeaway: What Does This Mean for Medicine?

This paper solves a long-standing debate. For decades, scientists argued whether the "Proton Sponge" (swelling) or "Membrane Fusion" (melting) was the key to escape.

The Verdict:

• Swelling alone isn't enough. The Classic Sponge (bPEI) is too weak to break out on its own.
• Grease is the key. To escape effectively, a delivery truck needs hydrophobic (oily) parts that can merge with the cell membrane.
• The Trade-off: The Greasy PBAE was the most effective at escaping, but because it melted the membrane so aggressively, it was also the most toxic (it hurt the cell). The LNP managed to do the same thing but more gently, which is why it's the safe, approved drug.

The Future:
The researchers are now saying: "If we want to design better drugs, we shouldn't just make stronger sponges. We should design polymers that have the right amount of 'grease' to melt into the cell wall, just like the successful LNP does, but without being too toxic."

In short: To escape a locked room, don't just push against the door. Find a way to melt the hinges.

Technical Summary

1. Problem Statement

Delivering nucleic acids (specifically siRNA) to the cytoplasm is a major bottleneck in gene therapy. After endocytosis, nanoparticles must escape the endo-lysosomal pathway to reach the cytoplasm and bind the RNA-induced Silencing Complex (RISC).

• The Gap: While various mechanisms have been proposed (e.g., the "proton sponge" effect, membrane fusion, pore formation), the specific molecular mechanisms driving efficient endosomal escape (EE) remain poorly understood.
• The Limitation: Existing experimental data (e.g., Gal8 recruitment assays) often fail to distinguish between effective escape and non-productive membrane damage or cytotoxicity. There is a lack of a predictive framework to fine-tune nanoparticle formulations for optimal EE without excessive toxicity.

2. Methodology

The study employs a multi-scale Molecular Dynamics (MD) simulation approach, correlating computational findings with in vitro experimental data.

• Simulation Techniques:
  • Coarse-Grained (CG) MD: Using the Martini 3 forcefield to simulate large systems (~100 nm) over long timescales (microseconds). This allowed for the simulation of nanoparticle self-assembly, interaction with planar membranes, and interaction with endosome-mimicking vesicles.
  • All-Atom (AA) MD: Using the CHARMM36 forcefield to provide high-resolution insights into specific binding mechanisms and transient interactions.
• Nanoparticle Formulations Simulated:
  1. bPEI: 25 kDa branched polyethylenimine (commercial standard).
  2. PPP: A block copolymer of 5 kDa bPEI linked by a 5 kDa polycaprolactone (PCL) chain.
  3. PBAE (30% OA): Amphiphilic poly(beta)aminoester with 30% hydrophobic oleylamine (OA) residues (more hydrophilic).
  4. PBAE (70% OA): Amphiphilic poly(beta)aminoester with 70% OA residues (more hydrophobic).
  5. LNP: An Onpattro®-like lipid nanoparticle (ionizable lipid MC3, DSPC, cholesterol).
• Membrane Models: Four distinct membrane models mimicking different stages of the endo-lysosomal pathway were constructed:
  • Early Endosome (slightly acidic, pH 6.5).
  • Lipid Raft (cholesterol/sphingolipid-rich).
  • Late Endosome (acidic, pH 5.4, high anionic lipids like BMGP).
  • Lysosome (acidic, pH 5.4, no glycolipids).
• Experimental Validation:
  • In vitro assays in HeLa cells: eGFP knockdown efficiency, cellular uptake (confocal microscopy), endosomal escape markers (Gal8 recruitment), cytotoxicity (LDH/CCK-8), and hemolysis assays.

3. Key Contributions

• Mechanistic Differentiation: This is the first study to visualize and distinguish the specific EE mechanisms of different polyplex formulations versus LNPs using MD simulations.
• Hydrophobicity vs. Charge: The study establishes that while electrostatic interactions with anionic lipids are necessary for initial attachment, hydrophobic interactions are the critical driver for effective membrane disruption and escape.
• Limitations of Gal8 Assay: The work demonstrates that Gal8 recruitment (a common EE marker) is not a universal predictor of therapeutic efficacy. LNPs showed high knockdown efficiency with low Gal8 recruitment, suggesting they escape via small, non-damaging pores rather than large membrane ruptures.
• Predictive Power: The MD simulations successfully predicted the in vitro stability and performance ranking of the formulations, validating MD as a tool for screening nanoparticle designs.

4. Key Results

A. In Vitro Performance

• Top Performers: The 70% OA PBAE polyplex and the LNP achieved the highest eGFP knockdown (30–60%).
• Poor Performers: bPEI, PPP, and 30% OA PBAE showed negligible knockdown.
• Cytotoxicity: The 70% OA PBAE caused significant cytotoxicity and LDH release, correlating with massive endosomal disruption. LNPs were non-toxic.
• Uptake vs. Escape: While 70% OA PBAE had superior cellular uptake, the LNP had comparable uptake to bPEI/PPP but superior knockdown, indicating the LNP's advantage lies in escape efficiency, not uptake.

B. MD Simulation Findings

• Interaction Mechanisms:
  • Hydrophobic Disruption (70% OA PBAE & LNP): These particles engaged in strong hydrophobic interactions with the lipid tails of the membrane. They extracted lipids from the membrane plane, caused significant membrane disordering, and facilitated lipid mixing. The LNP specifically underwent membrane fusion, leading to rapid lipid exchange and vesicle disruption.
  • Electrostatic Attachment (bPEI, PPP, 30% OA PBAE): These particles primarily attached to the membrane surface via electrostatic interactions with anionic lipids (e.g., BMGP). They caused minimal membrane disturbance and failed to extract lipids or induce fusion.
• Role of Anionic Lipids: All particles showed a strong preference for anionic lipids (BMGP, glycolipids). However, without hydrophobic residues, this attachment did not translate to escape.
• Proton Sponge vs. Fusion:
  • bPEI/PPP: The simulations supported a "proton sponge-like" mechanism where particles attach firmly, but escape relies on rare, high-energy events (osmotic/mechanical stress) that were not frequently observed in equilibrium simulations.
  • 70% OA PBAE: Exhibited a hybrid mechanism, combining electrostatic attachment with hydrophobic membrane fusion/disruption.
• Stability: The 70% OA PBAE showed the strongest membrane interaction but did not show the most efficient siRNA unpacking in the simulation, suggesting that membrane disruption and cargo release are distinct steps. The 30% OA PBAE was the least stable polyplex.

C. Correlation

• The number of polymer-membrane contacts in MD simulations correlated directly with in vitro knockdown efficiency and hemolysis.
• The 70% OA PBAE and LNP had 18–50x more membrane contacts than bPEI, explaining their superior performance.

5. Significance and Conclusion

• Paradigm Shift: The study challenges the exclusive reliance on the "proton sponge" theory for polymeric carriers. It posits that effective EE for polyplexes requires a spectrum of mechanisms, ranging from pure electrostatic attachment (inefficient) to hydrophobic-driven membrane fusion (highly efficient but potentially toxic).
• Design Guidelines: To improve siRNA delivery, polymer formulations should be engineered to include hydrophobic residues (like oleylamine) that facilitate membrane fusion and lipid extraction, similar to the mechanism of LNPs.
• Safety vs. Efficacy: While hydrophobic polyplexes (70% OA PBAE) are highly effective, they cause large membrane defects leading to cytotoxicity. LNPs achieve high efficacy through a "softer" fusion mechanism that creates small, non-lethal pores.
• Future Outlook: MD simulations are validated as a powerful tool to predict EE mechanisms and screen formulations, potentially reducing the need for extensive trial-and-error in vitro testing. The authors suggest future work should focus on engineering polyplexes that mimic the LNP's "fusion-based, low-toxicity" escape mechanism.